Win/Loss Pattern Analysis

Your ICP should not be a theory. It should be a statistical pattern extracted from your actual wins and losses. Pull every closed-won and closed-lost opportunity from the last 18 months. For each one, capture the firmographic attributes: industry, employee count, revenue range, technology stack, growth stage, and geography. Then look for the pattern. Which combination of attributes shows up disproportionately in wins? Which combination shows up disproportionately in losses? The gap between those two patterns is your ICP.

1. Pull the Raw Data Export closed-won and closed-lost opportunities with full account attributes from your CRM. You need at least 50 wins and 50 losses for the patterns to be meaningful. If you have fewer, extend the time window until you reach minimum sample size.
2. Cluster the Wins Group your wins by shared attributes. You will typically find two to three clusters — segments where your solution wins consistently. One cluster might be mid-market SaaS companies between 200-500 employees. Another might be financial services firms going through digital transformation. These clusters are your ICP segments.
3. Contrast with Losses Run the same clustering on your losses. The attributes that appear in losses but not wins are your anti-ICP — the accounts you should actively exclude from targeting. Excluding the anti-ICP is often more valuable than refining the ICP because it eliminates wasted cycles on accounts that will never close.